Appl Entomol Zool (2012) 47:379–388 DOI 10.1007/s13355-012-0130-x

ORIGINAL RESEARCH PAPER

Prediction of overseas migration of the small brown , striatellus (: ) in East Asia

Akira Otuka • Yijun Zhou • Gwan-Seok Lee • Masaya Matsumura • Yeqin Zhu • Hong-Hyun Park • Zewen Liu • Sachiyo Sanada-Morimura

Received: 19 April 2012 / Accepted: 25 July 2012 / Published online: 21 August 2012 Ó The Japanese Society of Applied Entomology and Zoology 2012

Abstract A method has been developed for predicting the Keywords Laodelphax striatellus Á Long-distance overseas migration of Laodelphax striatellus (Falle´n) from migration Á Effective accumulated temperature Á eastern China to Japan and Korea. The method consists of Migration simulation two techniques: estimation of the emigration period in the source region and simulation of migration. The emigration period was estimated by calculating the effective accumu- Introduction lated temperature for the by use of real-time daily surface temperatures at the source. During the emigration The small brown planthopper Laodelphax striatellus period, migration simulations were performed twice a day, (Falle´n), a vector of Rice stripe virus and Rice black at every dusk and dawn. The prediction method was eval- streaked dwarf virus, is one of the major insect pests of uated, by cross-validation using migrations in the 4 years rice. Because it is able to overwinter in temperate regions from 2008 to 2011. The results showed that the emigration including Japan, Korea, and China, most L. striatellus have periods included the mass migrations, and that the method long been believed not to be a migratory insect pest, successfully predicted those migrations. although small numbers of L. striatellus migrants have been caught over the sea (Noda 1986). Therefore, rice stripe disease in rice fields was believed to be caused mainly by local L. striatellus populations. Mass overseas migration of L. striatellus and a subsequent outbreak of rice A. Otuka (&) Á M. Matsumura Á S. Sanada-Morimura stripe disease were reported to have occurred in western NARO/Kyushu Okinawa Agricultural Research Center, Japan in 2008, and Jiangsu province in China was identi- 2421 Suya, Koshi, Kumamoto 861-1192, Japan fied as a possible source (Otuka et al. 2010). After this e-mail: [email protected] migration, another mass immigration occurred in western Y. Zhou Korea in 2009 (Kim 2009; Kim et al. 2009; Otuka 2009). Institute of Plant Protection, Jiangsu Academy of Agricultural Large numbers of immigrants were captured in net traps Sciences, Xuanwu, Nanjing 210014, People’s Republic of China along the western coastline of the Korean peninsula, and G.-S. Lee Á H.-H. Park the immigrants subsequently caused a severe outbreak of Crop Protection Division, National Academy of Agricultural rice stripe disease in paddy fields there (Kim 2009; Kim Science, Rural Development Administration, 61 Seodun-dong, et al. 2009). A possible source of these immigrations was Gwonseon-gu, Suwon 441-853, Republic of Korea estimated to be Jiangsu province (Otuka 2009). Y. Zhu Since 2004 a severe outbreak of L. striatellus and rice Plant Protection Station of Jiangsu Province, Longjiang, stripe disease has spread throughout Jiangsu province. Nanjing 210036, People’s Republic of China Viruliferous L. striatellus in the province peaked at 30 % in 2005 and decreased to 12 % in 2009 (Zhu 2006, 2009). Z. Liu College of Plant Protection, Nanjing Agricultural University, This was higher than the value of approximately 5 % for No. 1 Weigang, Nanjing 210095, People’s Republic of China Kyushu district, Japan (Matsumura and Otuka 2009). 123 380 Appl Entomol Zool (2012) 47:379–388

It has, thus, become important to predict the migration 126.34°E; open square in Fig. 1), Sinan, Korea (34.82°N, of L. striatellus to enable prediction of the pest’s arrival 126.10°E; open triangle), Isahaya, Japan (32.84°N, time and areas of invasion. 130.03°E; open circle), and Tongzhou, China (32.11°N, A migration-prediction technique has been applied to 121.08°E; solid triangle). Monitoring in Korea and China migration of two other tropical species of rice planthop- was started in 2009 and 2010, respectively. At each site, a pers—the brown planthopper, Nilaparvata lugens (Sta˚l), tow net trap 1.5 m in depth with a 1-m ring was mounted at and the white-backed planthopper, Sogatella furcifera the top of a pole 10 m above the ground. caught in (Horva´th)—and has achieved a prediction accuracy of each trap were collected every morning and counted by 78 % (Otuka et al. 2005a). In the prediction of migration extension officers, except that insects in the net trap at of these species, it is assumed that they take off in Taean in 2009 were collected every week until June 3 and southern China every dusk and dawn, mainly during the those in the same trap in 2011 were collected every 12 h at Bai-u rainy season from late June to early July, and their 0900 and 2100 hours local time on May 31 and afterward. overseas migration is predicted to occur when south- Insects in the net trap at Sinan in 2009 were not collected westerly winds blow over the East China Sea. Overseas on Saturdays and Sundays. migrations of L. striatellus have so far been confined to within very short periods of time, from late May to early Backward trajectory analysis June, as shown below. This period of emigration corre- sponds environmentally to the time of the wheat harvest To determine whether a catch in the net traps in Korea and and entomologically to the emergence of the first gener- Japan was a result of immigration from overseas, backward ation macropterous adults of L. striatellus (Zhu 2009, trajectory analysis was conducted (Otuka et al. 2005b). The Sogawa 2005). Therefore, prediction of migration on the analysis traces an air parcel backwardly to find a possible basis of the previously developed simulation technique migration source. The method has previously been applied was expected to be improved if the emigration period was to net trap catches obtained on 5 June 2008 in Isahaya and accurately estimated and the prediction was performed 1–3 June 2009 in Taean and Sinan, and those catches were during that period only. found to be a result of immigration from overseas (Otuka This study, therefore, proposes a method for prediction et al. 2010; Otuka 2009). The same analysis was applied to of migration involving a combination of the prediction of monitoring data in Korea and Japan in 2010 and 2011. The the emigration period of L. striatellus in the source region results were used for evaluation of the prediction. and simulation of migration. The former prediction is conducted by calculating the effective accumulated tem- perature (EAT) for the insect. The simulation model for tropical rice , with a few modifications, is used for the migration simulation. The prediction accuracy was evaluated by comparing the prediction with insect catches 40˚N in the destination areas. Meteorological conditions which cause the overseas migrations and possible domestic migrations within China are discussed. Taean

Materials and methods 35˚N Sinan

This section describes monitoring methods in three coun- Dongtai tries and outlines how to predict an overseas migration. Tongzhou Isahaya The method of prediction consists of two steps: prediction of the timing of emergence of the first generation of the overwintering population in a source region, and simula- 30˚N tion of migration. The section also describes a method for 115˚E 120˚E 125˚E 130˚E evaluation of a prediction. Fig. 1 Location of net traps in China, Korea, and Japan. The solid Trap monitoring in China, Korea, and Japan triangle indicates the location of the net trap at Tongzhou, Jiangsu province for emigration monitoring. The open square, triangle, and circle indicate net traps for immigration monitoring at Taean and To monitor the occurrence of L. striatellus, net traps were Sinan, Korea, and Isahaya, Japan, respectively. The solid circle located in three different countries: Taean, Korea (36.76°N, indicates the location of a weather station in Dongtai 123 Appl Entomol Zool (2012) 47:379–388 381

Migration source in this study

A whole region of Jiangsu province was assumed to be a a: EAT at emergece source area of L. striatellus overseas migration (the loca- tion of the province is given in Fig. 1). This assumption Past years was based on the following: – First, the province was estimated to be the possible source of migration into western Japan in 2008 (Otuka et al. 2010). b: Averaging termperature (EAT) – Second, L. striatellus has been widespread throughout Effective accumulated the whole of Jiangsu province since 2004 and its Date density in wheat fields in the spring has been very high (Zhu 2006, 2009). The area of occurrence of rice stripe c: Previously determined threshold disease in Jiangsu province accounted for 70 % of the Presumed daily increase total in China in 2003 (Zhou 2009).

EAT Prediction year An overview of predicting the timing of emergence and emigration Actual daily increase The timing of emergence and that of emigration are both of Date Predicted emergece date (PE) interest. Emigration occurs within a few days after emergence (Zhang et al. 2011). To predict the timing of emigration of PE L. striatellus from a source region, first, the EAT value at Date emergence of the first generation of the overwintering pop- ulation must be determined in advance; this calculation is d: Predicted emigration priod repeated for several previous years so that several EAT values Fig. 2 Schematic diagram of prediction of the emigration period. for the emergence are obtained (Fig. 2a). Second, these EAT The emergence date is predicted by calculating the effective values are averaged to obtain a previously determined accumulated temperature (EAT), and its threshold value is determined threshold. The threshold is used for prediction of the date of in advance by averaging the EAT values at emergence from past emergence for the current prediction year (Fig. 2b). years. The emigration period is determined by using the predicted emergence date (PE), taking into account both the yearly variation of In the prediction, the actual daily increase of the EAT EAT in the source region and the age of the emigrating insects (cf. value is calculated at one site in the source region, and section ‘‘The emigration period for prediction’’) presumed daily increases are added to the current EAT prediction’’ in ‘‘Results’’ below. The details of the EAT value until the sum goes beyond the previously determined calculation are presented in the next section. threshold (Fig. 2c). By use of this process, the date of emergence is determined. Effective accumulated temperature at emergence In addition, uncertainties in the EAT calculation must be considered, first because Jiangsu province is a large region To estimate the EAT value at emergence, accurate emer- and there may be deviations in the timing of emergence gence and/or emigration dates must be known (Fig. 2a). within the province as a result of different local tempera- The emergence and emigration dates in the source area tures. Another factor that introduces uncertainty into the were determined directly on the basis of the maximum calculation is the variation of the insect’s age when it emi- catch peak in the net trap located in southern Jiangsu in grates. During the emigration season in early June in 2009 2010, or indirectly via both immigration monitoring using and 2010 in Shandong province, the age of L. striatellus net traps in Korea and Japan and the estimated migration samples collected in a light trap had been estimated to be 1 duration over the sea in 2008, 2009, and 2011. The indirect to 3 days after emergence, on the basis of the development determination was performed as follows: of their ovaries (Zhang et al. 2011). Therefore, emigration should be considered to occur within a period of time that (Emigration date in Jiangsu) = (immigration date cap- includes the emergence date estimated above (Fig. 2d). The tured by the traps in Korea or Japan) - (migration migration simulations are performed during the emigration duration of 1 day) period. The length of the period was set to 9 days, for rea- (Emergence date in Jiangsu) = (emigration date in sons explained in the section ‘‘The emigration period for Jiangsu) - (pre-emigration period of 2 days)

123 382 Appl Entomol Zool (2012) 47:379–388 where the duration of migration was determined by the threshold (Fig. 2), and this date consequently determined analysis of the 2008 migration (Otuka et al. 2010), and the emigration period for the purpose of predicting the found to be 1 day. The pre-emigration period was deter- migration. During the emigration period, migration simu- mined by reference to Zhang et al. (2011), in which it was lation was performed to calculate the movement of L. stri- reported that L. striatellus adults 1–3 days old were caught atellus. The prediction results were evaluated by in a light trap during an emigration period. comparison with monitoring data. Because the number of The daily increase in EAT for the Jiangsu population overseas migrations of L. striatellus that had occurred by from 2008 to 2011 was calculated by use of the daily the time this paper was prepared was small, only three maximum and minimum surface temperatures at a station in migrations in 4 years (see ‘‘Results’’), cross-validation had Dongtai city (32.85°N, 120.28°E; solid circle in Fig. 1), to be performed to evaluate the validity of the whole pre- Jiangsu province. The station was selected because it is diction method, including prediction of emigration period located almost at the center of the province. The data were and prediction of migration. obtained from a database of the National Oceanic and In the cross-validation, first, a 4-year temperature data Atmospheric Administration, USA (WMO Resolution set from 2008 to 2011 was divided into a training set 40-NOAA), and the triangle method was used to calculate containing data for 3 years and a validation set containing EAT (Sakagami and Korenaga 1981) with a developmental data for 1 year. For example, three EAT values for 2008, zero temperature of 11 °C, a developmental maximum 2009, and 2010 accumulated from January 1 to the esti- temperature of 29 °C, and a developmental arrest tempera- mated emergence date of the first generation were initially ture of 40 °C. The first two values were determined by determined with a training data set. The three values were reference to Noda (1989). The last value was used so that the then averaged, and the average, as a threshold, was used to developmental arrest would not be taken into account, predict the emergence date in 2011. because the developmental arrest temperature is not known A predicted EAT value for each day was calculated both and the daily maximum temperature in early June in the with real-time daily temperatures until the present day and source region is usually not very high. The EAT value at with an assumed daily EAT value of 11 degree days each growth stage is listed in Table 1, and a value of beyond that day. 374.3 degree days for one generation was used. The cal- Likewise, three other combinations of the training set, culation was started on 1 January each year and terminated (2009, 10, 11), (2008, 10, 11), and (2008, 09, 11), were on the estimated emergence date of the first generation of the produced, and the averaged EAT values were calculated overwintering population (estimated above). The first day of and used to predict the emergence dates for 2008, 2009, the year was used because no growth was expected during and 2010, respectively. the winter and, therefore, it is an appropriate starting point. Model for prediction of migration Prediction of migration, and its evaluation The migration simulation model was based on a model for The emergence date was determined as a date when the the brown planthopper, Nilaparvata lugens (Fig. 3; Otuka EAT of L. striatellus exceeds a previously determined et al. 2005a). However, three modifications were made for L. striatellus. First, a take-off square, only, had been used previously; in this work a polygonal take-off area was used, Table 1 The effective accumulated temperature (EAT) to each so that a take-off area with a complicated shape such as growth stage of the small brown planthopper, Laodelphax striatellus that of Jiangsu province could be used. Second, the take-off time used to be set manually at dusk or dawn (Otuka et al. Stage EAT (degree days) 2005a); this was modified to be automatically calculated on the basis of the sun’s orbit (Nagasawa 1999), thus enabling One generation 374.3 accurate setting of take-off times in a wide take-off area Pre-oviposition 54.3 with a complex shape. Hatching 110.0 Last, the temperature ceiling was modified (Fig. 3). The 2nd instar 150.5 temperature ceiling is a level across which simulated 3rd instar 183.1 planthoppers do not enter a higher atmospheric region with 4th instar 218.6 lower air temperatures. For N. lugens, the temperature 5th instar 258.3 ceiling was set to 16.5 °C, at which half of N. lugens Emerging 320.0 stopped beating their wings in a tethered flight experiment The EAT value for pre-oviposition is from Noda (1989) and the under laboratory conditions (Ohkubo 1973). For L. stria- others for the later stages are from Hachiya (1997) tellus, the ceiling was set to 13 °C, at which the new 123 Appl Entomol Zool (2012) 47:379–388 383

Take-off Flight Over Japan

temp. ceiling sim. duration 24 h

13°C

horizontally random take-off at dawn or dusk

upward velocity 0.2 m/s relative aerial density for 1 h wind velocity vartical diffusion

L. striatellus 100 m

33 km grid take-off area

Jiangsu prov., China East China Sea Western Japan

Fig. 3 Schematic diagram of the method of simulation. A dark a ceiling level with a temperature of 13 °C. Arriving over Japan ellipse represents an insect. At take-off (left), small brown planthop- (right), the number of insects is counted in each calculation grid cell pers take off from the slanted belt area where the sun sets or rises. The to produce the relative areal density, whose non zero areas at the insects move upwards for 1 h after take-off. During the flight (center), lowest level (ground surface to 100 m), which are called migration they move at the same velocity as winds, taking vertical diffusion into clouds, are mapped. Figures drawn by use of this method are shown in account. They do not enter upper levels higher, or regions cooler, than Fig. 6

migration simulation model well reproduced the 2008 The catch at Isahaya resulting from immigration from overseas migration in a preliminary examination. overseas is that made on June 5 2008 (Fig. 4a; Otuka et al. Because no landing process of the insects is considered 2010). Small catches from May 29 to June 1 were a result in the model because of lack of knowledge, all simulated of emigration from the local wheat fields, which were planthoppers can fly until the simulation stops. A predicted being harvested (Otuka et al. 2010). duration of 24 h was therefore used to avoid overestimated In 2009, there was no catch at Isahaya. A large catch, mal-predictions. however, was made on June 1 at Sinan, Korea (Fig. 4b). Prediction of migration is made up two steps: weather Because this trap was not operational on Saturdays and prediction and prediction of migration. Weather was pre- Sundays, the possible immigration date ranges from May 30 dicted by use of a numerical weather-prediction model, to June 1. The net trap located at Taean was operated weekly, MM5 (NCAR 2003), with both global gridded weather and had a large catch of 963 insects on June 3. Backward data, i.e. GSM data of the Japan Meteorological Agency trajectory analysis suggested that the catch at Sinan was a (JMA 2007), and sea surface temperature data, i.e. result of immigration from Jiangsu province (Otuka 2009). RTG_SST of the US National Oceanic and Atmospheric In 2010, no obvious catch was observed in Korea or Administration (Thiebaux et al. 2001). Predicted wind and Japan from late May to early June. Backward trajectory air temperature data were used by the modified migration analysis showed that easterly winds blew over Korea and simulation model that predicted the relative aerial density western Japan during early June (data not shown), sug- of the insect. Non-zero regions of the relative aerial density gesting that no immigration from overseas occurred in are called ‘‘migration clouds’’. Korea and Japan. There is a catch peak of 972 on June 10 at Tongzhou (Fig. 4c) which indicates a large emigration. In 2011, the first catch occurred at Taean, Korea during Results the night of May 31 (from 2100 hours on May 31 to 0900 hours on June 1) (Fig. 4d), and a subsequent catch Monitoring data occurred in the next 12 h and lingered for an additional 12 h. Later than that, interestingly, outstanding catches were The catches of L. striatellus in the net traps in Japan, observed only during the day time. No insect was caught Korea, and China are shown in Fig. 4. during the night. As for 2008 and 2009, backward trajectory

123 384 Appl Entomol Zool (2012) 47:379–388

70 1000 a b 963 Isahaya 2008 900 Sinan 2009 60 Isahaya 2010 Taean 2009 819 800 50 700 600

40 at Taean, Korea at Isahaya, Japan 500 30 400

20 300 L. striatellus

L. striatellus 200 10 132 100 31 0 0 1 6 6 3 0 of Catch 0 Catch of Catch 9 11 Jun 1 Jun 3 Jun 5 Jun 7 Jun Jun 1 Jun 2 Jun 3 Jun 4 Jun Jun May 16 May 18 May 20 May 22 May 24 May 26 May 28 May 30 May 27 May 30 May 28 May 29 May 31

1000 c 500 Tongzhou 2010 d 900 450 Taean 2011 800 400 700 350 600 300 at Taean, Korea

at Tongzhou, China 500 250 400 200 300 150 L. striatellus 200 100 L. striatellus 100 50

0 of Catch 0 2 Catch of Catch 10 14 30 DT NT NT NT NT NT Jun Jun 4 Jun 6 Jun 8 1 2DT Jun Jun 12 Jun Jun 16 Jun 18 May 31 May 29 May 27 May 29 May Jun Jun Jun 3 DT Jun 4 DT Jun 5 DT May 31 DT

Fig. 4 Net trap monitoring in China, Korea and Japan. The respectively. The Japanese data were obtained from the database abbreviations DT and NT stand for monitoring period in the daytime service of the Japan Plant Protection Association (JPP-NET 2012) (0100–1300 hours UTC) and nighttime (1300–0100 hours UTC), analysis suggested that the catches made during the night on As an example, trends of the EAT value in 2008, 2009, May 31 and during the daytime on June 1 were a result of and 2010 are shown in Fig. 5b. The EAT values did not immigration from Jiangsu province (data not shown). The increase much during January and February, and started to terminal points of the backward trajectories of 24-h travel grow similarly in March in each of the 3 years, but they then duration for the catches made in the daytime on June 2 and 3 diverged, because of subsequent different temperature trends were distributed over the sea. It is suggested those catches in the spring. The EAT prediction for May 30 2011 is shown were not a result of immigration from overseas. in Fig. 5a. The latest value was that on May 28 (2 days before May 30) because of the time lag of the data distri- EAT values from 2008 to 2011 bution. The EAT threshold, an average of the EATs in 2008, 2009, and 2010 in Table 2, was set at 453.7 degree days, The EAT values from January 1 to the emergence date each which was exceeded by the predicted EAT on June 2. year are shown in Table 2. The emigration peak in 2010 was directly observed in daily catches in the net trap in The emigration period for prediction Tongzhou, China. The emergence dates in 2008, 2009, and 2011 were estimated from immigration dates in Japan and Table 2 indicates the variations in the EAT value at Korea, which were monitored by use of the net traps there. emergence (variations of -39.1 to ?35.9 degree days from

123 Appl Entomol Zool (2012) 47:379–388 385

Table 2 Effective accumulated temperature for L. striatellus from because L. striatellus adults 1–3 day(s) old after emergence January 1 to emergence date have been observed to emigrate from Shandong province, Year Effective Emergence Emigration Immigration China (Zhang et al. 2011), and the 3 days of error in the accumulated date date in the date in Japan emergence prediction plus 2 days for the pre-emigration temperature source or Korea make up a total of 5 days. In total, the emigration period 2008 476.2 June 2 June 4 June 5 in western for prediction of migration was set at 9 days (3 days ? Japan 1 day of emergence date ? 5 days). 2009 445.4 May 29 May 31 June 1 in western Korea Prediction of migration, and its evaluation 2010 438.4 June 8 June 10 (No immigration occurred.) For example, an EAT threshold for emergence prediction 2011 401.2 May 28 May 30 May 31 in in 2011 was found to be 453.3 degree days (Table 3). By western Korea use of this value, on 30 May 2011 the emergence date was predicted to be June 2 (Fig. 5a). The predicted migration started on May 30 (=June 2—3 days) and continued until Jun 2 June 7 (Table 3). Examples of predicted migration clouds a Prediction 453.7 degree•day 2011 are shown in Fig. 6. Among a total of 18 predictions in 2011, three that started from Jiangsu province on May 30 and 31 were to arrive over western Korea on May 31 and June 1 (Fig. 6g–i), corresponding to catches during the May 28 night on May 31 and during the day on June 1 (Fig. 4d). The prediction of migrations to Korea were therefore successful. A prediction of migration to Japan on 6 June 2011 was incorrect, because no corresponding catch was observed in southern Kagoshima (JPP-NET 2012). The predicted area and the evaluation of the predictions for 4 years in the cross-validation are summarized in Table 3 and Fig. 6. It is apparent that the emigration period b 2008 of 9 days included all overseas migrations in 2008, 2009, 2009 and 2011. Though there was no overseas migration in 2010 2010, the emigration peak on June 10 in China was included in the emigration period. In some predictions, migration clouds moved to North Korea (N. Korea in Table 3), where no monitoring data were available and the result was unknown. Outside the emigration periods, it is believed that no overseas migration occurred in the 4 years studied. The overall hit ratio for predictions (= the number of successful predictions/the total number of predictions) excluding the unknown cases was found to be 0.93 (29/31) (Table 3). In addition, to compare the emergence dates in Table 2 that were assumed to be the true values, with the dates Fig. 5 Effective accumulated temperature (EAT) for the Chinese L. striatellus population. a shows the prediction of EAT in 2011. The predicted on Day 0 in Table 3, the error of the emergence predicted value exceeded the threshold of 453.7 degree days on June prediction was found to be within -3to?5 days of the 2, which was estimated on May 30. b shows the trend of EAT values emergence dates. in 2008, 2009, and 2010. The calculation was terminated on the estimated emigration date in each year Discussion the average). These values corresponded to -3.6 to ?3.3 days if a daily increase of 11 degree days was used. This paper presents a method for prediction of migration of Because of this error, a time of 3 days was used for L. striatellus consisting of the prediction of the emigration pre-emergence in the emigration period. A time of 5 days period and migration simulation during that period. The was selected for post-emergence in the emigration period emigration period was predicted by use of the EAT, which 123 386 Appl Entomol Zool (2012) 47:379–388

Table 3 Cross-validation of the prediction of overseas migrations Year EAT threshold Prediction date (degree days) Day -3 Day -2 Day -1 Day 0a Day ?1 Day ?2 Day ?3 Day ?4 Day ?5 Predicted immigration area: prediction resultb

2008 428.3 May 27 May 28 May 29 May 30 May 31 Jun 01 Jun 02 Jun 03 Jun 04 Korea: u NM: S NM: S NM: S N. Korea: u NM: S NM: S NM: S Japan: S Japan: 9 2009 438.6 May 26 May 27 May 28 May 29 May 30 May 31 Jun 01 Jun 02 Jun 03 NM: S NM: S NM: S NM: S Korea: u Korea: S N. Korea: u NM: S NM: S 2010 440.9 Jun 06 Jun 07 Jun 08 Jun 09 Jun 10 Jun 11 Jun 12 Jun 13 Jun 14 NM: S NM: S NM: S NM: S NM: S NM: S NM: S NM: S NM: S 2011 453.3 May 30 May 31 Jun 01 Jun 02 Jun 03 Jun 04 Jun 05 Jun 06 Jun 07 Korea: S Korea: S NM: S N. Korea: u NM: S NM: S NM: S Japan: 9 N. Korea: u N. Korea: u a Day 0 of the prediction date corresponds to predicted emergence on Day -3 b A prediction is successful (S) when a predicted immigration area (Fig. 6) and a catch of immigration there (Fig. 4) matched each other, or when no predicted migration (NM) and no catch occurred concurrently in Korea or Japan. Otherwise, the prediction is wrong (9). Letter u represents unknown result because of the lack of monitoring data in Korea or North Korea (N. Korea) was calculation from developmental data for Japanese the 4th instar has previously been reported as a main dia- L. striatellus populations and surface temperatures pause stage (Kisimoto 1958), the estimated 5th nymph in observed at the Chinese weather station in the center of 2011 is more advanced than that. The emigration date in Jiangsu province. The method of prediction was evaluated 2011 was also earlier in the emigration period than in other by the cross-validation method and net trap data in the years (Table 2). In the previous study, southern Japanese immigration countries. The results showed that the over- L. striatellus populations emerged over shorter periods than seas migrations in 2008, 2009 and 2011 were included in a northern one under laboratory conditions (Noda 1992). the emigration periods and were successfully predicted as This suggests there might be a similar difference in the simulated migration clouds, indicating that the proposed diapause characteristics of the Chinese populations. It is method is effective. Prediction of migration for future important to examine emigration timing and area in the years will be performed with an EAT threshold of source area in relation to the insect’s diapause. 440.3 degree days, an average of the four EAT values in Interestingly, catches from June 2 to 5 in 2011 at Taean Table 2. occurred only during the daytime (Fig. 4d). Although the The reason why the EAT value in Table 2 has decreased catch on June 3 was the largest in Fig. 4d, there was no year by year is not clear. There may be regional differences prediction of migration from Jiangsu to Taean. Addition- in the insect’s development in the source because Jiangsu ally, the backward trajectory analysis found no other province is a very large area, yet the temperature data for Chinese regions as a source (data not shown). These day- the prediction came from one weather station only in the time catches might, therefore, be a result of a local popu- central province. Division of the source into smaller lation, or of relocation activities of immigrants that arrived regions might, therefore, improve prediction of the emi- from China by June 1. However the origin of the catches is gration period and overseas migration; this should be fur- still unknown. ther investigated. The weather conditions necessary to bring L. striatellus Another topic related to the EAT is the growth stage of into Japan in early June were investigated in a previous the L. striatellus Jiangsu population on the first day of the study (Syobu et al. 2011). It was found that cold vortexes year. Because emigrating insects in early June are those of were the driving force for the overseas migrations. The the first generation of the overwintering population, the frequency of this phenomenon, however, has been esti- growth stage can be inversely estimated by subtracting an mated to be low, occurring only twice in the most recent EAT value of 374.3 degree days for one generation 10 years (2000–2009). On the other hand, in Korea, the (Table 1) (Noda 1989; Hachiya 1997). Both the remaining frequency of overseas migrations may be larger than in values, ranging from 101.9 to 26.8 degree days, and EAT Japan, currently occurring at least twice in 4 years values at various stages in Table 1 suggest that the hiber- (2008–2011). The outbreak of the disease also occurred in nating stage on January 1 was the 3rd to 5th instar. Because the western coastal regions of the Korean Peninsula in

123 Appl Entomol Zool (2012) 47:379–388 387

a b c

40˚N

30˚N 11:00 UTC May 28 21:00 UTC May 28 11:00 UTC May 5

120˚E 130˚E 140˚E 120˚E 130˚E 140˚E 120˚E 130˚E 140˚E 11:00 UTC May 27 2008 21:00 UTC May 27 2008 11:00 UTC May 4 2008

d e f

40˚N

30˚N 21:00 UTC May 31 11:00 UTC June 1 21:00 UTC June 1

120˚E 130˚E 140˚E 120˚E 130˚E 140˚E 120˚E 130˚E 140˚E 21:00 UTC May 30 2009 11:00 UTC May 31 2009 21:00 UTC May 31 2009

g h i

40˚N

30˚N 11:00 UTC May 31 21:00 UTC May 31 11:00 UTC June 1

120˚E 130˚E 140˚E 120˚E 130˚E 140˚E 120˚E 130˚E 140˚E 11:00 UTC May 30 2011 21:00 UTC May 30 2011 11:00 UTC May 31 2011

j k l

40˚N

30˚N 21:00 UTC June 1 21:00 UTC June 3 21:00 UTC June 7

120˚E 130˚E 140˚E 120˚E 130˚E 140˚E 120˚E 130˚E 140˚E 21:00 UTC May 31 2011 21:00 UTC June 2 2011 21:00 UTC June 6 2011

Fig. 6 Examples of prediction of migration. Migration clouds (dark areas) after 24 h (at the upper time) from each take-off time (lower italic time) are shown. Areas on the land covered with migration clouds are the predicted immigration areas

2001, 2007, and 2008 although their causes have not yet The prediction technique for L. striatellus migrations been analyzed (Kim 2009). This possible difference in developed in this study is mainly for overseas migrations frequency between the destinations may be due to differ- from China to Korea or Japan. The same method may be ences in their direction and distance from the source. Korea applicable to analysis and prediction of domestic migra- is closer to Jiangsu province than western Japan, and tions in China. In fact, for example, migration clouds in southwesterly winds are typical when a low pressure sys- 2010 moved to the north or northwest from Jiangsu prov- tem occurs in the northern hemisphere. A study similar to ince, suggesting that domestic migrations occurred. It has that by Syobu et al. (2011) is preferred to determine the also been reported that the outbreak of rice strip diseases frequency of overseas migrations in Korean cases. has spread to adjacent provinces, such as Zhejiang, Anhui,

123 388 Appl Entomol Zool (2012) 47:379–388

Shandong, and Hubei (Wei et al. 2009; Wang et al. 2008). Otuka A (2009) Migration analysis of the small brown planthopper, Analysis of domestic migrations is a topic for a future Laodelphax striatellus, in western Japan and Korea. In: Proceedings of APEC workshop on the epidemics of migratory study. insect pests and associated virus diseases in rice and their impact on food security in APEC member economies, Seoul, Acknowledgment We thank Dr Gu-Feng Zhang for net trap mon- pp 145–157 itoring of the small brown planthopper at Tongzhou Plant Protection Otuka A, Watanabe T, Suzuki Y, Matsumura M, Furuno A, Chino M Station, Jiangsu province, China. Mr Jeong Tae-Woo at Taean (2005a) Real-time prediction for migration of rice planthoppers Agricultural Development and Technology Center, Chungnam prov- Sogatella furcifera (Horva´th) and Nilaparvata lugens (Sta˚l) ince, Korea is also acknowledged for net trap monitoring of the insect. (Homoptera: Delphacidae). Appl Ent Zool 40:221–229 Otuka A, Dudhia J, Watanabe T, Furuno A (2005b) A new trajectory analysis method for migratory planthoppers, Sogatella furcifera References (Horva´th) and Nilaparvata lugens (Sta˚l) (Homoptera: Delphaci- dae). Agric For Entomol 7:1–9 Otuka A, Matsumura M, Sanada-Morimura S, Takeuchi H, Watanabe Hachiya K (1997) Studies on the population dynamics and the pest T, Ohtsu R, Inoue H (2010) The 2008 overseas mass migration Laodelphax management of the small brown planthopper, of the small brown planthopper, Laodelphax striatellus, and striatellus ´ Fallen. Rep Hokkaido Pref Agric Exp Stn 90:1–77 subsequent outbreak of rice stripe disease in western Japan. Appl JMA (2007) Global Spectral Model (JMA-GSM0603). In: Outline of Entomol Zool 45:259–266 the operational numerical weather prediction at the Japan Sakagami Y, Korenaga R (1981) ‘Triangle Method’, a simple method Meteorological Agency, Japan Meteorological Agency, Tokyo, for the estimation of total effective temperature. Jpn J Appl Ent pp 41–66 Zool 25:52–54 (in Japanese with English summary) JPP-NET (2012) The database service of the Japan Plant Protection Sogawa S (2005) Epidemic of rice stripe virus disease in Jiangsu Association, Tokyo. http://www.jppn.ne.jp. Accessed 1 Apr province, China. Nogyo Gijutsu 60:405–409 (in Japanese) 2012 Syobu S, Otuka A, Matsumura M (2011) Trap catches of the small Kim JS (2009) Forecasting and occurrence of SBPH and RSV in brown planthopper, Laodelphax striatellus (Falle´n) (Hemiptera: Korea. In: Proceedings of APEC workshop on the epidemics of Delphacidae), in northern Kyushu district, Japan in relation to migratory insect pests and associated virus diseases in rice and weather conditions. Appl Entomol Zool 46:41–50 their impact on food security in APEC member economies, Thiebaux J, Katz B, Wang W (2001) New sea-surface temperature Seoul, pp 89–106 analysis implemented at NCEP. WRCP series report no. 115, Kim CS, Lee GS, Choi HS (2009) Virus-insect-plant interaction at WMO/TD-no. 1083 RSV outbreak regions in Korea. In: Proceedings of APEC Wang HD, Chen JP, Zhang HM, Sun XL, Zhu JL, Wang AG, Sheng workshop on the epidemics of migratory insect pests and WX, Adams MJ (2008) Recent Rice strip virus epidemics in associated virus diseases in rice and their impact on food security Zhejiang province, China, and experiments on sowing date, in APEC member economies, Seoul, pp 61–74 disease-yield loss relationships and seedling susceptibility. Plant Kisimoto R (1958) Studies on the diapause in the planthoppers, effect Dis 92:1190–1196 of photoperiod on the induction and the completion of diapause Wei TY, Yang JG, Liao FL, Gao FL, Lu LM, Zhang XT, Li F, Wu ZJ, in the fourth larval stage of the small brown planthopper, Lin QY, Xie LH, Lin HX (2009) Genetic diversity and ´ Delphacodes striatella Fallen. Jpn J Appl Ent Zool 2:128–134 population structure of rice stripe virus in China. J Gen Virol Matsumura M, Otuka A (2009) Recent occurrence of the small brown 90:1025–1034 planthopper and the rice strip virus disease in Japan. Plant Prot Zhang HY, Diao YG, Yang HB, Zhao Y, Zhang XX, Zhai BP (2011) 63:293–296 (in Japanese) Population dynamics and migration characteristics of the small Nagasawa K (1999) Computations of sunrise and sunset. Chijinsho- brown planthopper in spring in Jining, Shandong province. Chin kan, Tokyo (in Japanese) J Appl Entomol 48:1298–1308 NCAR (2003) PSU/NCAR mesoscale modeling system tutorial class Zhou YJ (2009) SBPH and RSV interactions. In: Proceedings of the note and user’s guide: MM5 Modeling System version 3. MMM APEC workshop on the epidemics of migratory insect pests and National Center for Atmospheric Research, Boulder associated virus diseases in rice and their impact on food security Noda H (1986) Pre-mating flight of rice planthopper (Homoptera: in APEC member economies, Seoul, pp 27–54 Delphacidae) collected on the East China Sea. Appl Entomol Zhu YQ (2006) The occurrence of small brown planthopper (SBPH) Zool 21:175–176 and controlling techniques of the Rice stripe virus (RSV). In: Noda H (1989) Developmental zero and total effective temperature of Proceedings of the international workshop on ecology and three rice planthoppers (Homoptera: Delphacidae). Jpn J Appl management of rice planthoppers, Hangzhou, China, pp 16–17 Ent Zool 33:263–266 (in Japanese) Zhu YQ (2009) Status of occurrence and control for SBPH and RSV Noda H (1992) Geographic variation of nymphal diapause in the in Jiangsu province. In: Proceedings of the APEC workshop on small brown planthopper in Japan. Jpn Agric Res Q 26:124–129 the epidemics of migratory insect pests and associated virus Ohkubo N (1973) Experimental studies on the flight of planthoppers diseases in rice and their impact on food security in APEC by the tethered flight technique. Jpn J Appl Ent Zool 17:10–18 member economies, Seoul, pp 15–25 (in Japanese with English summary)

123